Würstchen
Efficient latent diffusion model using highly compressed latent space.
About
Wurstchen is a text-to-image latent diffusion framework that moves the expensive text-conditional stage into a highly compressed latent space. It adds a second compression stage so that Stage A and Stage B reconstruct images while Stage C learns the text-conditional model in a low-dimensional space, reaching a 42x compression factor and reducing training and inference cost while keeping image quality. Created by Pablo Pernias under the MIT license.
Reviews (0)
Leave a Review
No reviews yet. Be the first to review!
Details
- Category
- Image Generation
- Price
- Free
- Platform
- Local/Desktop
- Difficulty
- Intermediate (3/5)
- License
- MIT
- Minimum VRAM
- 4 GB
- Added
- Apr 3, 2026
Related Tools
Optimized Stable Diffusion WebUI fork focused on faster inference and easier development.
Efficient text-to-image diffusion transformer producing 4K resolution images.
Next-generation image generation model by Black Forest Labs.
Feature-rich web interface for Stable Diffusion image generation
Text-to-image model by Sber AI with latent diffusion architecture.
Simplified Stable Diffusion interface inspired by Midjourney